ITU-T Kaleidoscope 2009 Innovations for Digital Inclusion

1. QoS Management for ISP: A Model and Implementation Methodology based on ITU-T Rec. E.802 Framework ITU-T Kaleidoscope 2009Innovations for Digital Inclusion Eva Ibarrola University of the Basque Country (UPV/EHU) eva.ibarrola@ehu.es

2. Outline • Introduction • Background • Proposal • QoS management model • Deployment methodology • Case study • Results • Conclusions

3. Introduction • New broadband services • Competitive condition • QoP differential factor among ISPs • From “best effort” to “user centered” • Quality of service (E.800): • “… its ability to satisfy stated and implied needs of the user of the service.” • Appropriate QoS policies required

4. Background • ITU-T QoS Framework: Rec. G.1000 • Four points of view • Regulator’s approach • Difficult to deploy Methodology needed! • QoS management process: Rec. E.802 • QoS procedures • Well detailed • Theoretical stage Strategy needed!

5. QoS Management Proposal • QoS Model • QoS Criteria • Methodology • Satisfaction • Expectation

6. QoS Models Accesibility Continuity Fullfilment Customer Care Network QoS Availability • ITU-T Rec. E.802: • Universal Model • Performance Model • Four Market Model • ACF model • CSAT Model Need for simplification Network QoS Availability Customer Care Content

7. QoS Management Model • Identification of QoS criteria • Considered on a service by service basis • Specified on an end-to-end basis • Different population segments may have different QoS preferences • QoS profile for customers are time dependant • A set of functions to cover the QoS criteria • Measurable indicators/parameters • QoS perceived can be modeled

8. QoS Model FUNCTION A FUNCTION B SERVICE A SERVICE C SERVICE B PQoS B ExpQoS A EXPQoS A PQoS B EXPQoS B EXPQoS B EXPQoS B PQoS A PQoS A PQoS C PQoS B EXPQoS C PQoS A EXPQoS A MEASURABLE OBJECTIVE OR SUBJECTIVE PARAMETERS FUNCTION A NETWORK QoS FUNCTION C FUNCTION D NETWORK AVAILABILITY FUNCTION E FUNCTION A CUSTOMER CARE FUNCTION B FUNCTION H FUNCTION A CONTENT QoS FUNCTION J

9. Methodology Proposal Satisfaction Model Customer’s QoS Requirements QoS Offered By Provider Expectation Customer’s Expectation Customer’s Satisfaction Expectation Update Disconformation QoS Perceived By Customer QoS Achieved By Provider Methodology Four Market Model QoS criteria Performance Model Universal Model ACF Model CSAT Model ITU-T E.802 QoS Model Customer’s Perception Framework SERVICE PROVIDER CUSTOMER ITU-T G.1000

10. Case Study • Web service • Residential users • Sample date results from experiment • Steps: • Identification of user’s requirements • Specification of functions and parameters • Customer satisfaction analysis

11. User’s QoS Requirements • Network QoS criteria • Maximum download and upload speed • Availability criteria • Availability of web service • Customer Care criteria • Premium service support • Content criteria • No gathered data

12. Functions and Parameters Functions Session Time Service Availability Service Support • Parameters Throughput Percentage of time service available Response time for service support

13. QoS offered • Maximum throughput • Upload • Download • Maximum Availability • Availability • Best response time • For premium service support: • Defective Service Instance (DSI): • Notion to quantify the negative effect of offering beyond user’s requirements

14. QoS Delivered • Measured throughput • Average upload: • Average download: • DSI penalization factor for session time: • Measured availability • Percentage of time considered available: • Total access time: • DSI penalization factor for availability: • Measured average response time • Average response time measured: • DSI penalization factor for support:

15. QoS Perceived by User User’s Preferences γ1, γ2 – download and upload (γ1 + γ2 = 1) α1, α2, α3 – Network QoS, availability and customer care (α1 + α2 + α3 = 1) Session Time Availability Service Support Perceived Utility

16. Customer Satisfaction • Customer Satisfaction: • Expectation: • Disconformation: • Perception function: • Disconformation function:

17. Experiment • Carried out on June 2008 on Web service • Two kind of users: expert and unskilled • Different network conditions (delay) • Capture their preferences: • Satisfaction evaluation

18. Results User’s overall satisfaction closely linked to contextual parameters. User’s requirements change over the time and under different conditions. Different users have different requirements.

19. Conclusions (I) • The QoS model and implementation methodology grants a lot of precision to the ITU-T QoS Framework • The results of the experiment demonstrate the usefulness of our approach • Users, ISPs and regulators may find a lot of benefits when deploying the QoS management model

20. Conclusions (II) • The QoS management model and deployment methodology totally fulfills international standards • The proposal could be used to complete the specification of QoS management processes in future standards or revisions • Even applicable to NGN

21. Thank you!Any questions? ITU-T Kaleidoscope 2009Innovations for Digital Inclusion Eva Ibarrola University of the Basque Country (UPV/EHU) eva.ibarrola@ehu.es